A question of Noise

[3dbeing]

According to the Houdini 10.* Docs

Original Perlin noise is similar to Perlin noise, but marginally less efficient in computation and with different characteristics. The bounds on the noise are roughly (-1, 1).

I would assume this to imply that the bounds for both are (-1,1)

However when I use Perlin noise in a turbulent noise vop my output bounds are roughly (.33,.65)

Using the Original Perlin option I get bounds of (-.3,.35)

I emailed SESI, but the answer was quite obfuscated; so I thought someone here might be able to clear it up... What do i need to do, not counting a fit vop, but rather what do I need to do to get a rough bounds of (-1,1) out of the turbulent vop?

Cheers,

-=3db

vop_turbulent_noise_v1.hip.gz

[Macha]

Yes, it's odd this. It obviously depends on the input positions but it should at least go negative and it never does. It's interesting but unfortunately I have no time at the moment to pursue this.

Anyway to help debug this I attached a setup that visualizes the problem. Left-click on the attribute value.

vop_turbulent_noise_v1.hipnc

[johner]

Original Perlin noise is similar to Perlin noise, but marginally less efficient in computation and with different characteristics. The bounds on the noise are roughly (-1, 1).

I would assume this to imply that the bounds for both are (-1,1)

However when I use Perlin noise in a turbulent noise vop my output bounds are roughly (.33,.65)

Using the Original Perlin option I get bounds of (-.3,.35)

Here's the H10 docs for the Perlin noise VEX function. The Turbulent Noise VOP uses this noise function for Perlin noise, which has a range from 0 to 1. Your test sphere is not large enough nor does it have high enough density points to see the whole range of noise, although it's actually unlikely you'll ever see anything over .9 or less than .1.

I'm not sure why the range of onoise (Original Perlin) is limited, but you're probably better off with Perlin anyway. I would just pipe the output to a Fit Range VOP, with the Dest Min and Max set to -1 and 1.

One thing to be aware of: since turbulent noise is generated by adding multiple octaves of noise according to the Turbulence parameter, you can end up with values greater or less than the range of the underlying noise function itself.

[3dbeing]

Thanks for the file Macha excellent idea visualizing the output in a graph.

Johenr, I see now if I increase the size and the resolution of the sphere to a dramatic level I do get a bounds of roughly (0,1). I guess the docs are a bit misleading.

Thanks for the help guys...

[old school]

All of these noise functions are clearly defined in the Advanced RenderMan and the Procedural Texturing and Modeling books.

No mystery with noise functions after you read those books.

Btw, you can implement the noise functions in COP VOPs and visualize the noise there. Inspect all the various noise functions to see min/max ranges and more.

[MENOZ]

All of these noise functions are clearly defined in the Advanced RenderMan and the Procedural Texturing and Modeling books.

No mystery with noise functions after you read those books.

Btw, you can implement the noise functions in COP VOPs and visualize the noise there. Inspect all the various noise functions to see min/max ranges and more.

Hi,

the "Procedural Texturing and Modeling" book is this one?

Texturing and Modeling, Third Edition: A Procedural Approach (The Morgan Kaufmann Series in Computer Graphics)

http://www.amazon.com/Texturing-Modeling-Third-Procedural-Approach/dp/1558608486